September 8, 2026
Summary
This article explains the methodology used to generate provincial rankings of CVITP coverage of adults living in poverty. Because Statistics Canada does not publish annual provincial counts of adults living in poverty, the analysis combines two official datasets: provincial population estimates by age, and provincial Market Basket Measure (MBM) poverty rates. To isolate the adult population, the 15–19 age group in the population table is apportioned so that only the 18–19 portion is included in the adult count. For each province and year, the estimated number of adults living in poverty is calculated by multiplying the adult population by the MBM poverty rate. The territories are excluded from this exercise as their MBM poverty rates are generally not reliable or absent.
These provincial poverty estimates are then compared with CRA’s annual CVITP statistics, which report the number of adult clients helped in each province. Dividing CVITP adult clients by the estimated number of adults living in poverty gives a coverage rate for each province.
Because the provincial poverty estimates are biased downward, absolute coverage rates cannot be compared across years. However, the bias isidentical across provinces and consistent across years. Therefore, while the absolute values are not precise, the relativedifferences between provinces are real and relevant.
Overview
This article provides a detailed explanation of the methodology used (with justification) for the rankings generated in the article titled “Inequalities of Access Between Provinces to Free CVITP Services: Annual Rankings (2021-2025)“.
To compare provincial CVITP performance, I take annual estimates of adults living in poverty in each province and compare these with CRA‑published CVITP client counts. Because Statistics Canada does not publish annual provincial counts of adults living in poverty, this method combines two datasets to produce internally consistent provincial estimates that allow for cross‑provincial comparisons.
This analysis covers the ten provinces only. Statistics Canada’s Market Basket Measure (MBM) poverty estimates for the territories are incomplete and often carry high sampling‑variability flags (E or F), indicating that the underlying survey data are too sparse for reliable inference. Several territorial age‑group poverty rates are missing entirely, and others are statistically unstable to the point where producing population‑level poverty counts would be misleading. Territorial data are therefore excluded.
Introduction
Three datasets are used as the basis for this analysis, the first two generated by Statistics Canada:
1Population estimates by age and province: Table 17‑10‑0005‑01 (Population estimates on July 1, by age and sex)
Why use population estimates on July 1 rather than year-end estimates?[i]
2Provincial poverty rates by age group (MBM): Table 11‑10‑0135‑01 (Low income statistics by age, gender and economic family type)
3CRA CVITP statistics by province
Step 1 — Calculating the number of adults by province
To compare CVITP coverage across provinces, I first need provincial estimates of adults (18+) living in poverty.
Statistics Canada’s population estimates by age come in five-year bands (0-4, 5-9, 10-14, etc). I need population estimates for under 18 years of age, 18-64 and 65+.
First, I need to apportion the 15–19 age group which straddles the 18-year threshold which constitutes the year when one can file a return, regardless of one’s income level, and is thus considered an adult.
To isolate the 18+ population, I apportion the 15–19 group assuming a uniform distribution across ages:
- 3/5 of the 15–19 group allocated to ages 15–17
- 2/5 allocated to ages 18–19
The 18–19 estimate is then added to the 20+ population to produce a provincial estimate of adults between 18 and 64. This apportionment method ensures consistency across provinces and years. Finally, the 65+ group is easy to calculate as 65 is the beginning of the five-year band that runs from 65 to 69 and I simply add all the age bands from 65 and higher together.
Step 2 — Estimating the number of adults living in poverty by province
The provincial MBM poverty rates are given for three separate age groups: under 18, 18-64 and 65+. Rates are also given a score for reliability. E and F are the least reliable and as many of the territories age group ratings get these scores, or have no score at all, I exclude the territories from this exercise.
For each province and each year, the following calculation is made:
Adult population (18-64) × Adult (18-64) MBM poverty rate
And
Adult population (65+) x Adult (65+) MBM poverty rate
These two numbers are then summed to produce an estimate of the number of adults living in poverty. This method is necessary because Statistics Canada does not publish annual provincial poverty counts for adults; the Canadian Income Survey (CIS) only provides national totals.
Why are provincial poverty estimates generally lower than CIS national totals?[ii]
Step 3 — Calculating the CVITP coverage rates by province
The CRA publishes annual CVITP statistics, including the number of clients helped by province. I make two assumptions: all CVITP clients are living in poverty and at least 18 years of age or older.
The first assumption relies on the fact that most host organizations select clients based on the CRA’s suggested income ceilings. These income ceilings take the household structure into account. While they make use of total income and MBM poverty thresholds are based on disposable income, my experience from analyzing data in a small number of tax clinics suggests that over 90% of clients served are living below the MBM poverty threshold.
As to the second assumption, I rely again on my experience from working in a number of tax clinics to infer that the number of clients served who are under 18 is negligible.
For each province and each year, I calculate:
CVITP coverage rate = CVITP clients ÷ estimated adults in poverty
These coverage rates form the basis of the annual provincial rankings. Finally, I compare a province’s annual rankings.
Why can rankings between years be compared but not coverage rates?[iii]
[i] The July 1 dataset is the only one that provides the age‑specific detail required to isolate the adult population. Although mid‑year estimates differ slightly from year‑end counts, the difference is very small and does not meaningfully affect provincial comparisons. More importantly, any minor bias introduced by using July 1 estimates applies uniformly across all provinces and all years. This ensures that while absolute coverage rates cannot be compared across years, the provincial rankings remain valid.
[ii] The population × MBM rate method underestimates the true number of adults living in poverty because:
- MBM rates are rounded to one decimal place
- mid‑year rather than year end population estimates are being used
- absence of microdata: The Canadian Income Survey uses microdata — individual level survey records that allow Statistics Canada to apply family level poverty rules, regional MBM thresholds, income imputation, non response adjustments, and calibrated survey weights. (These refinements cannot be replicated using provincial population estimates and published MBM rates.)
- poverty data from the territories is not included
- poverty data for Indigenous populations living on reserve is also excluded
These limitations affect all provinces equally, which is crucial for the legitimacy of the rankings.
[iii] Because the provincial poverty estimates are biased generally downward, absolute coverage rates cannot be compared across years. However, the bias is identical across provinces and consistent across years. Therefore, while the absolute values are not precise, the relative differences between provinces are real and relevant.